Seventeen studies (5307 patients) were included. Of these 12 were randomised controlled trials (RCTs) and 5 were controlled studies.
The methodological quality was poor, with only 3 of the 17 studies scoring over 50 (out of 100). Four methodological problems were prevalent, specifically accounting for co-interventions, lack of placebo group, blinding of assessment of outcome, and analysis not based on 'intention to treat'.
Analysis by type of aftercare programme showed that the majority of studies did not report clear beneficial effects for the intervention group. When aftercare programmes were compared to usual care (8 studies) or the reference treatment was no special nursing care (6 studies), most effects were not beneficial for the intervention group. In the 3 studies comparing intervention with institutional care, positive effects were evident for quality of care and costs.
Overall, the analysis of outcome measures showed that studies with negative or mixed outcomes had similar methodological scores, but studies with only positive results showed lower methodological scores. For the five outcome measures: none of the 10 studies assessing quality of life had positive outcomes; 3 studies of patient compliance, 2 with negative outcomes had higher methodological quality than studies with positive outcome; 6 studies of costs, 5 studies with positive effect for intervention had higher methodological scores than negative studies; 10 studies of medical consumption, 7 studies with negative effects for intervention had higher methodological scores than positive studies; and 2 studies of quality of care both had positive effects for intervention.
Comparison of the RCTs with control trials showed that only 2 of the 12 RCTs had positive outcomes, whilst all 7 controlled trials had positive effects on one or more outcomes.